Ventricle Contact Is Associated With Higher 11C Methionine PET Uptake in Glioblastoma

Introduction: Ventricle contact is associated with a worse prognosis and more aggressive tumor characteristics in glioblastoma (GBM). This is hypothesized to be a result of neural stem cells located around the lateral ventricles, in the subventricular zone. 11C methionine Positron Emission Tomography (metPET) is an indicator for increased proliferation, as it shows uptake of methionine, an amino acid needed for protein synthesis. This study is the rst to study metPET characteristics of GBM in relation to ventricle contact. Methods: 12 patients with IDH wild type GBM a were included. Using MRI, the following regions were determined: primary tumor (dened as contrast enhancing lesion on T1) and peritumoral edema (dened as edema visible on FLAIR excluding the enhancement). PET parameters in these areas were extracted using PET fused with MRI imaging. Parameters extracted from the PET included: maximum and mean tumor-to-normal ratio (TNRmax and TNRmean) and metabolic tumor volume (MTV). Results: TNRmean of the primary tumor showed signicantly higher values for the ventricle contacting group compared to the non-contacting group (4.44 vs 2.67, p=0.015). Other metPET parameters suggested higher values for the ventricle contacting group, but these differences did not reach statistical signicance. Conclusion: GBM with ventricle contact show a higher methionine uptake and thus increased proliferation compared with GBM without ventricle contact. This might explain survival differences and should be considered in treatment decisions.


Introduction
Glioblastoma (GBM) is a high-grade primary brain malignancy with a dire prognosis. Despite adequate treatment median survival ranges from 12 to 18 months [1,2]. This short life expectancy, is related to the almost certain recurrence of GBM, which is due to the in ltrative nature of the tumor [3][4][5]. Recently, it was demonstrated that GBM with ventricle contact have a worse prognosis when compared to GBM without contact. A meta-analysis showed that ventricle contact is associated with lower survival independent of other prognostic in uences [6].
The subventricular zone is an important neural stem cell containing brain region and hypothesized to play a role in the tumorigenesis of GBM [7,8]. Although ventricle contact has not yet been shown to be related to a speci cally different cell lineage in this regard, it has been associated with more therapy resistance [9]. In addition, ventricle contacting GBMs have been shown to be more often multifocal and recur further away from the primary tumor, both features being associated with poorer survival [6,10]. These results seem to suggest that, although a difference in cell of origin has not yet been demonstrated, ventricle contact is associated with more aggressive tumor behavior. We hypothesized that because of their location, ventricle contacting GBM could be more prone to stem cell in uence and take on more aggressive characteristics, including increased proliferation, than non-contacting GBMs.
MR imaging, the most frequently used imaging modality in studies of GBM, provides several indirect indicators of proliferation, such as tumor size and contrast enhancement. However, neither of these parameters give a direct indication of biological activity of the tumor and are thus limited in appreciating aggressiveness and proliferation.
11C methionine PET (metPET), a widely employed amino acid tracer, has been shown to indirectly re ect tumor proliferation, as methionine uptake is associated with protein synthesis [11][12][13]. However, to our best knowledge, studies evaluating metPET characteristics of GBM in relation to ventricle contact have not yet been performed. We hypothesize that a more aggressive behavior of ventricle contacting GBM will be shown by a higher proliferation indicated by metPET.

Patient population
Data for this study were acquired retrospectively from patient les in the University Medical Centre of Groningen (UMCG) in the Netherlands from 2011-2019. The study has been approved by the institutional review board and the need for written informed consent was waived. We included patients with isocitrate dehydrogenase wildtype (IDHwt) GBM, following the most recent EANO (European Association of Neuro-Oncology) guidelines [14]. Patients were only eligible for inclusion if they conformed to the following inclusion criteria: 1) IDHwt GBM was con rmed by pathological report, 2) preoperative/prebiopsy metPET imaging was available and 3) preoperative MRI with at least T1 post-contrast and T2 FLAIR sequences was available. The MRI was required to assess the location of the tumor and to determine the peritumoral area. Exclusion criteria included 1) no contrast enhancing lesion on MRI, 2) other neurological malignant processes and 3) age below 18. Included patients were rst scored on whether the tumors showed ventricle contact by BS and BVD independently, after which an inter-rater agreement score was calculated using a Cohen's Kappa score [15]. Con icting scores were settled by AVDH, a neuro-radiologist with more than 5 years of experience in neuro oncology. Ventricle contact was determined by the extent of the contrast enhancing lesion on the T1 post-contrast sequence on MRI. This was in line with previous studies [16,17]. The date of diagnosis was de ned as the day of the biopsy or operation that con rmed the diagnosis via pathological examination.

MRI and PET acquisition
Both MRI and PET scans were extracted from the medical les of the included patients. MRI imaging was performed on several different types of machines, from different manufacturers. All MRI were performed on 1.5T or 3.0T. The imaging protocols at least included a 3D post-contrast T1 sequence (repetition time [TR] 2000-2250 msec, echo time [TE] 2.67-3.40 msec, inversion time [TI] 850-900 msec, slice thickness 1 mm, no slice gap, voxel size 1x1x1 mm) and uid-attenuated inversion recovery (FLAIR) sequence (TR 5000-11000 msec, TE 87-337 msec, TI 1800-2800 msec, slice thickness 1-5 mm, no slice gap). Most protocols also included axial T1 pre-contrast, T2, DWI. Models that were used for scanning included Philips ingenia, Siemens aera, Siemens sonata and Siemens avanto. All PET imaging was performed in the UMCG, using either a Siemens Biograph mCT (N = 10) or PET-HR+ (N = 3) scanner (Siemens, Knoxville, Tennessee. Prior to the PET scan, patients fasted for at least six hours. Static imaging was performed 20 to 40 minutes after intravenous injection of 11C methionine was. For the mCT camera, a low dose CT was acquired for attenuation correction and images were reconstructed using Truex + TOF with 3 iterations and 21 subsets in a 400 x 400 matrix size (zoom 1.0). For the HR + camera, a transmission scan was performed immediately after emission scanning in the same bed positions in order to correct for attenuation and images were reconstructed using OSEM with 3 iterations and 21 subsets. The mean dosage was 205 MBq (range 192 to 224 MBq), radiochemical purity was always higher than 95% and the speci c activity higher than 10000 GBq/mmol. The 2006 EANM procedure guidelines for brain tumor imaging using labeled amino acid analogues were followed [18].

Image analysis
The different volumes of interest (VOI), including contrast enhancing tumor and peritumoral edema were delineated manually. This delineation was performed on the available preoperative MRI using 3DSlicer version 4.10.2 (http://www.slicer.org). The primary tumor was delineated as the contrast enhancing area on T1, internal cystic or necrotic tissue was not excluded from the primary tumor VOI. Peritumoral area was determined using the T2 FLAIR sequence, it was de ned as the area of high intensity surrounding the tumor, excluding the primary tumor VOI as already determined on the post-contrast T1 weighted sequence. These delineations were made manually on a slice by slice basis as illustrated on Fig. 1A&B. Besides these sequences, unenhanced T1 and T2 were also used to support assessment. The determined VOIs were then used as a supporting tool in PET analysis using PMOD version 4.1 (PMOD Technologies, Zurich, Switzerland). PET scans were co-registered to corresponding MRI, making it possible to use the MRI for overlays on the PET. Using the semiautomatic delineation tools in PMOD, the primary tumor as visible on T1 post-contrast weighted MRI was converted into a mask, which could be overlaid on the PET scan (Fig. 1C&D). These VOIs were compared to the predetermined delineations of 3DSlicer and adjusted where necessary. Additionally, the high tracer uptake PET lesion was also delineated semi automatically using PMOD software. The mask was used to divide the high PET uptake lesion into two parts: 1) the high PET uptake lesion inside the mask and 2) the high PET uptake lesion outside the mask.
PET parameters PET parameters were collected from these VOIs separately, where parameters from inside the mask were de ned as those of the primary tumor and parameters from outside the mask were de ned as those of the peritumoral area. Several PET parameters were determined within these two VOIs and included maximum and mean tumor-to-normal-ratios (TNRmax and TNRmean, respectively) as metabolic parameters and metabolic tumor volume (MTV) as volumetric parameter. For each voxel, the standardized uptake value (SUV) was determined as follows.

SUV = C(T)/[injection dose(MBq)/patient weight (kg)] [19]
The TNR values were calculated by dividing the corresponding SUV value within the VOI by the SUV value within a 1-cm radius sphere mirrored on the contralateral centrum semiovale. MTV represents the volume of the PET enhancing area in mL.

Statistics
All statistical analysis was performed in SPSS version 23.0 (Armonk, NY, IBM Corp.). Patient characteristics were compared using the corresponding non-parametric statistical tests. Proportions were compared using a chi square test and medians were compared with a Mann Whitney U test. Median values of the PET parameters and tumor volumes were determined and differences in rankings were tested using the Mann Whitney U test. Statically signi cance was set at a two-sided p-value for the general patient characteristics. As we hypothesized more aggressive features in ventricle contacting tumors, we used a one-sided p value of 0.05 for these analyses.

Patient Population
The 12 included patients had a median age of 59 years (range 41-73). No signi cant differences were observed between the two groups. The characteristics and associated signi cance values are shown in Table 1. After scoring these patients for ventricle contact, there was only one disagreement resulting in a good inter-rater agreement with a kappa of 0.80 (20). Of the 12 included patients, 5 were found to have a ventricle contacting tumor (38.5%), while 7 had a non-contacting tumor (61.5%).

11C methionine PET Parameters
All PET parameters of both the primary tumor VOI and the peritumoral VOI demonstrated higher median values for the ventricle contacting group compared to the non-contacting group, as well as higher mean ranks (Tables 2 and 3

Discussion
This is the rst study comparing metPET characteristics of GBM in relation to ventricle contact. Our results show an increased methionine uptake for ventricle contacting GBMs in comparison to noncontacting GBMs. Speci cally, our results demonstrated signi cantly higher TNRmean for ventricle contacting tumors than non-contacting tumors. This would imply an increased protein synthesis and therefore higher proliferation in the ventricle contacting group, further supporting the more aggressive characteristics of ventricle contacting GBMs.
Ventricle contact in GBM has previously been associated with several imaging characteristics related to tumor aggressiveness, such as larger tumor volumes, more multifocal and distant recurrences, and higher peritumoral perfusion, all of which potentially contribute to the lower survival of these tumors [6,9,10]. In addition to this, we have now also demonstrated a higher methionine uptake, as a measure of increased proliferation, for ventricle contacting GBM.
Although our study is the rst to study metPET in relation to ventricle contact in GBM, our results are in line with a recent study looking at cellular proliferation with O-(2-[18F] uoroethyl)-L-tyrosine (FET) PET [21]. Although it was not the primary research question, FET uptake showed a signi cantly higher SUVmean value for GBM with ventricle involvement when compared to GBM without ventricle contact.
We also hypothesized that ventricle contact in GBM could be related to higher methionine uptake levels in the peritumoral area. The peritumoral area plays an important role in the recurrence of GBM as it contains both vasogenic edema and in ltrating tumor cells. The peritumoral area is of particular interest in relation to ventricle contact in GBM as one characteristic of ventricle contacting GBM is a more distant and multifocal recurrence [6,10].
In addition, a recent study showed a higher peritumoral perfusion for ventricle contacting GBMs, associated with a more aggressive peritumoral in ltration [22]. In line with these hypotheses, we did nd higher values for metPET parameters in the peritumoral region, but none of those differences reached signi cance.
Despite the signi cant difference in TNRmean between ventricle contacting and non-contacting GBMs, this study does have its limitations. The rst limitation is the small sample size. This fact is a direct cause for the inability to meet assumptions for normal distribution and asymptotic signi cance, as well as a limited power of the outcomes. It is because of this that this study should primarily be seen as a pilot study and an invitation to perform larger prospective studies. Nevertheless, a signi cant difference for TNRmean has been observed, even with this limited sample size. Furthermore, the other PET parameters point in the same direction although not reaching signi cance. A second limitation is the retrospective nature of our data, resulting in the lack of a standardized protocol, such as the use of two PET systems with different resolutions. These different resolutions could potentially have introduced partial volume effects. However, by employing TNR and thus dividing uptake values by the background uptake, we aimed to limit these partial volume effects. The lack of standardization also resulted in a heterogeneous time interval between PET and MRI acquisition, which in some cases amounted to as much as 29 days. Prospective studies should make sure that time between MRI and PET is minimalized, which would lead to a smaller inaccuracy when performing the MRI overlay on the PET.

Conclusion
Ventricle contacting GBMs demonstrate a signi cantly higher uptake of methionine than non-contacting tumors. The increased uptake of methionine indirectly indicates a higher proliferation. These ndings possibly explain the survival difference between ventricle contacting and non-contacting GBMs and should be considered in treatment decisions.

Declarations
Funding: No funding was received for conducting this study.
Con icts of interest: The authors have no con icts of interest to declare that are relevant to the content of this article.
Availability of data and material: The data generated and analyzed during the current study are available from the corresponding author on reasonable request.